Genomic determinants of cisplatin resistance in germ cell tumors.

5027 Background: Cisplatin-based chemotherapy achieves complete remission in 70-80% of patients with germ cell tumors (GCTs), but a subset are cisplatin-resistant and progress after first-line therapy. Prior studies have linked TP53 and MDM2 alterations and chromosome 3p25.3 gain to cisplatin resistance and worse outcomes in GCT. We sought to validate this and identify additional genomic determinants of cisplatin response in a large GCT cohort. Methods: Patients who received first-line cisplatin-based chemotherapy for GCT of any primary site with genomic analysis of the tumor were eligible. Cisplatin resistance was defined as: (1) incomplete response to chemotherapy, (2) non-teratomatous progression after chemotherapy, or (3) viable non-teratomatous GCT at post-chemotherapy surgery; all other patients were considered sensitive. Next-generation sequencing was performed using the MSK-IMPACT assay and association between genomic alterations and resistance was assessed using Fisher’s exact test. Cytoband analysis was completed using the FACETS algorithm to calculate chromosomal alterations, with a required cytoband coverage of 75%. The association between cytoband alterations and resistance was assessed using relative risk with a robust variance estimator. Results: The cohort of 556 patients included 79% testicular, 15% mediastinal, 4% ovarian, and 1.5% tumors of other primary site. 79% were nonseminomatous germ cell tumors and 21% pure seminoma. Among these, 336 (60%) were cisplatin-resistant. TP53 mutations were more common in resistant than sensitive tumors (16% vs 2%), as were MDM2 and CRKL amplifications (7% vs 1% and 4% vs < 1%, respectively). KIT was more frequently mutated in cisplatin-sensitive tumors (5% in resistant vs 14% in sensitive) (Table 1). TP53 mutations occurred predominantly in mediastinal primary tumors (89%) whereas MDM2 alterations were confined to testicular primaries. Of 472 patients who underwent cytoband 3p25.3 analysis, chromosome 3p25.3 gain was associated with cisplatin resistance (relative risk 1.23; 95% CI: 1.00, 1.49; p = 0.04). Conclusions: In this large GCT cohort, TP53 and MDM2 alterations and chromosome 3p25.3 gains were associated with cisplatin resistance, consistent with prior reports. We also found that CRKL amplifications were associated with cisplatin resistance whereas KIT mutations were associated with cisplatin sensitivity. These latter findings are the first reports of the association of these genes with cisplatin response and warrant further mechanistic investigation to define how these alterations influence cisplatin sensitivity. Genomic alterations corresponding to cisplatin response in GCT. Alteration Resistant (n=336) Sensitive (n=220) p-value TP53 mutations 53 (16%) 4 (2%) <0.01 KIT mutations 16 (5%) 31 (14%) <0.01 MDM2 amplifications 23 (7%) 2 (1%) <0.01 CRKL amplifications 14 (4%) 1 (<1%) <0.01